My home server is dying unexplainedly. Totally locks solid, nothing in the logs, very confusing. While I figure out what’s wrong (power supply?) I decided to go implement a watchdog to reboot the system if it fails.
This turns out to be hard in 2019, with Ubuntu 19. tl;dr there are two choices: systemd or the good ol watchdog daemon. See also: softdog.
systemd has watchdog support. Unlike many things in the systemd hydra, this one makes sense to me to integrate. It’s configured in /etc/systemd/system.conf. However it’s not well documented and relatively limited, so I decided not to use it. Even so my system has a PID 77 named [watchdogd] that I think is part of systemd. No idea what it’s doing, if anything.
watchdog is the old school Linux daemon that does watchdoggy things. Ubuntu 19 offers version 5.15-2, and if you install it by default it…
View original post 568 more words